OPINION OF THE COURT
Whistleblowing does not insulate an employee from being fired for misconduct. Alastair Crosbie reported signs of possible fraud. More than a year later, his coworker accused him of harassment. His employer investigated the accusation and fired him within two days. Crosbie says flaws in the investigation show that it must have been a sham. He insists that the real reason he was fired was his whistleblowing, not his coworkers complaint. Because he has not debunked his employers explanation, we will affirm.
I. Background
Crosbie was hired by Gateway to help Highmark, a health-insurance company, investigate fraud. (Because Gateway wrote his paychecks but a Highmark employee decided to fire him, we call them both his employers.) While auditing Highmarks network of doctors, Crosbie found some troubling facts. For instance, he says he discovered that some doctors had prior convictions for selling opioid prescriptions. Others, he maintains, lacked required Medicaid licenses. In mid-2017, he reported his concerns to his managers at Gateway. They investigated but decided not to take any action. Yet Crosbie kept pressing the issue. Eventually, his managers told him to drop it.
Fast-forward to October 1, 2018, more than a year after his first report. Crosbies coworker lodged a complaint against him. She said that Crosbie had called her “Miss Piggy” and “oinked” at her. App. 455–56. Gateways human-resources team (HR) investigated. Their investigator interviewed Crosbie, the complainant, and an eyewitness who corroborated the complainants story. She also interviewed other people who knew of past issues between Crosbie and the complainant. And she spoke to Jim Burgess, one of the managers who had told Crosbie to drop the fraud issue. Burgess told the investigator that he would have questioned the allegations. But earlier that day, Crosbie had called him and made “coughing” and “snorting” noises, which made him think that Crosbie “did it.” App. 1076, 1313. On October 3, HR fired Crosbie.
Crosbie shot back, suing Gateway and Highmark under the False Claims Act for retaliation. Crosbie claimed that they had fired him because of his fraud reports. The employers replied that the people who had decided to fire Crosbie knew nothing about his reports and that they had good reason to fire him. Agreeing, the District Court granted summary judgment. Crosbie, it concluded, had not shown that the employers reason was a mere pretext for retaliation.
Crosbie now appeals. We review the grant of summary judgment de novo and draw every reasonable inference in Crosbies favor. Sikkelee v. Precision Airmotive Corp., 907 F.3d 701, 708 (3d Cir. 2018). Summary judgment is proper if “there is no genuine dispute as to any material fact.” Fed. R. Civ. P. 56(a). Here, there is not.
II. Crosbie Does Not Refute the Employers Reason for Firing Him
We normally analyze retaliation cases under a three-step burden-shifting framework. First, the employee alleges that he has been fired (or demoted or the like) for protected conduct. Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 142, 120 S.Ct. 2097, 147 L.Ed.2d 105 (2000). Then, the burden shifts to the employer to give a valid basis for firing him. Id.; see Hutchins v. Wilentz, Goldman & Spitzer, 253 F.3d 176, 186 (3d Cir. 2001) (False Claims Act retaliation). Last, the burden returns to the employee to show that the alternative reason is just “a pretext for discrimination” or retaliation. Reeves, 530 U.S. at 143, 120 S.Ct. 2097 (internal quotation marks omitted).
We have never held that this three-step framework governs False Claims Act claims. But the parties do not dispute it, so we assume without deciding that it does. We also assume, as did the District Court, that the parties have passed the first two steps. Because pretext is where the action is, we focus there.
Crosbie can prove pretext one of two ways. First, he can so thoroughly disprove his employers explanation for firing him that a jury could find it “unworthy of credence.” Fuentes v. Perskie, 32 F.3d 759, 765 (3d Cir. 1994) (internal quotation marks omitted). That means showing not just “that the employers proffered reason was wrong, but that it was so plainly wrong that it cannot have been the employers real reason.” Keller v. Orix Credit All., Inc., 130 F.3d 1101, 1109 (3d Cir. 1997) (en banc). Or Crosbie can cut to the chase and show directly that retaliation “was more likely than not a motivating or determinative” reason for his firing. Fuentes, 32 F.3d at 762. Either way, the question is not whether firing Crosbie was “wise,” but whether his employers retaliated. Id. at 765. And either way, he needs evidence. Id. at 762; see also Sheridan v. E.I. DuPont de Nemours & Co., 100 F.3d 1061, 1067 (3d Cir. 1996) (en banc).
Crosbie thinks he has enough evidence to support either pretext theory. First, he points to the quality of the harassment investigation. It was so flawed, he argues, that it must have been a sham. Second, he stresses Burgesss participation in that investigation. He alleges that Burgess, who knew of Crosbies protected activity, used the harassment investigation to get rid of him. In other words, Burgess used the investigation as a “cats paw.”
Either way, Crosbie lacks proof. His mere suspicions of shady behavior do not show pretext. We hold that Crosbie cannot show retaliation just by pointing out an investigations flaws. Nor can he prevail on his cats-paw theory without showing that the people who fired him relied on Burgesss statements.
A. An imperfect investigation alone does not show pretext
First, Crosbie objects that the investigation was slapdash. The investigator, he says, did not follow standard procedure, interview every witness, or look at the complainants history of baseless allegations. But sloppiness is not enough. The question is not whether Gateway conducted the “best, or even a sound” inquiry, but whether the investigation was a sham, a mere pretext to retaliate. Keller, 130 F.3d at 1109 (internal quotation marks omitted).
This investigation was far from a facade. HR received a complaint from Crosbies coworker. They interviewed her, Crosbie, and an eyewitness who corroborated the complainants account. At Crosbies urging, they also interviewed other witnesses who were aware of past issues between Crosbie and the complainant. Perhaps they could have done more. But Crosbie fails to show that the investigation was so thoroughly flawed that a jury could find it unbelievable. Cf. Fuentes, 32 F.3d at 765.
Crosbie next insists that even if he cannot show that the investigation was pretextual, its timing supports an inference of retaliation. It does not. True, we sometimes infer retaliation when someone is fired conspicuously soon after he blows the whistle. See, e.g., Farrell v. Planters Lifesavers Co., 206 F.3d 271, 279–81 (3d Cir. 2000). But Crosbie has the opposite problem. Much of the alleged protected conduct predated his firing by more than a year. Crosbie sent his last complaint at least a month before Gateway fired him. Yet he was fired a mere two days after he apparently oinked at his coworker. So the reasonable temporal inference is that Gateway fired him for oinking.
In any event, an investigations quality or timing can support an inference of pretext only if those running the investigation know of the protected activity. Crosbie concedes that HR did not know about his False Claims Act reports. So that team could not have been using the investigation to cover their tracks because they had no reason to think that they had tracks to cover. They, not upper management, decided to fire him. And because HR did not know of his reporting, it could not have fired him in retaliation. See Daniels v. Sch. Dist. of Phila., 776 F.3d 181, 196 (3d Cir. 2015) (“The plaintiff ․ cannot establish ․ a causal connection without some evidence that the [decisionmakers] knew of the plaintiffs protected conduct at the time they acted.”).
B. Crosbies cats-paw theory fares no better
This brings us to Crosbies second argument. Even if HR itself did not mean to retaliate, some nefarious manager might have used it as an instrument of retaliation—a cats paw. See Staub v. Proctor Hosp., 562 U.S. 411, 419, 131 S.Ct. 1186, 179 L.Ed.2d 144 (2011). Crosbies theory goes like this: HR got a harassment complaint and investigated it. At some point, they spoke to Burgess, who knew of Crosbies False Claims Act reports. Angry about those reports, Burgess wanted Crosbie out. So he seized on the opportunity, telling the investigator that he believed the harassment complaint. Relying on that comment, HR prematurely ended the investigation and fired Crosbie. Or so he says.
An employer can be liable on the cats-paw theory only if a non-decisionmakers act proximately caused the firing. Sometimes causation can be inferred from circumstantial evidence. Farrell, 206 F.3d at 280–81. But a bald assertion is not enough. Our cases emphasize three requirements for a cats-paw theory to succeed:
1. Retaliatory animus. The non-decisionmaker must be motivated by a desire to retaliate. See McKenna v. City of Philadelphia, 649 F.3d 171, 179 (3d Cir. 2011) (requiring proof that retaliatory animus played a “direct and substantial” and “foreseeable” role in an employees firing).
2. Communication. The non-decisionmaker must communicate with the decisionmaker. See Staub, 562 U.S. at 421, 422, 131 S.Ct. 1186 (concluding that a “biased report” can count as that “act”); McKenna, 649 F.3d at 179 (biased disciplinary record); Abramson v. William Patterson Coll. of N.J., 260 F.3d 265, 286 (3d Cir. 2001) (evaluations, memos, and comments).
3. Reliance. Then, in firing (or otherwise acting against) the employee, the decisionmaker must rely on the non-decisionmakers communication. See Jones v. Se. Pa. Transp. Auth., 796 F.3d 323, 331 (3d Cir. 2015).
Crosbies cats-paw theory fails at both the first and third prongs. On animus, he offers some disputed evidence that Burgess disliked him. But he has no evidence that Burgess wanted him fired, much less that he wanted him fired because of his False Claims Act reporting. After all, Burgess told the investigator that he would have questioned the allegations but for Crosbies behavior on their phone call about the investigation.
Crosbies reliance problem is worse. Even if he could show that Burgess meant to retaliate, he has no evidence that Burgesss discussion with the investigator influenced the investigation or his firing. That train had left the station long before Burgess got involved. The investigator started investigating Crosbie before she ever talked to Burgess. By the time she got to Burgess, she had questioned several witnesses, including Crosbie himself and the complainant. There is no reason to think that Burgesss comments caused the investigation to end early or changed the decisionmakers mind.
Speculation is no substitute for evidence. Because Crosbie has no evidence that Burgess played a causal role in his firing, summary judgment was proper.
III. Crosbie Cannot Show That Lack of Discovery Prejudiced Him
Last, Crosbie objects that the District Court refused to compel discovery of the complainants personnel file. Perhaps he does not have enough evidence now, he thinks, because he did not have the chance to get it. We review the District Courts denial for abuse of discretion. Anderson v. Wachovia Mortg. Corp., 621 F.3d 261, 281 (3d Cir. 2010). Crosbie must show that the denial caused him “actual and substantial prejudice.” Id.
Yet Crosbies proof of prejudice is speculative at best. He gives us little reason to think that extra discovery would have changed the outcome. True, the complainants personnel file could show that she had a history of making unfounded complaints about her coworkers. Even so, there still would not be enough evidence to show that HR was “plainly wrong” to believe her complaint. Keller, 130 F.3d at 1109. HR decided based on an investigation that involved multiple witnesses, including an eyewitness who corroborated the complainants story. Plus, even finding dirt on the complainant would at most suggest that she was lying, not that managers were using her complaint as a pretext to retaliate.
* * * * *
Crosbie says he was fired not because of his coworkers harassment complaint but because he had reported possible fraud. But he fails to show that the harassment investigation was a sham. Nor does he show that Burgess manipulated the investigation to get him fired in retaliation. So we will affirm.
BIBAS, Circuit Judge.